The West Delphi Group
A Proposal for Case Definitions and Outcome Measures in Studies of Infantile Spasms and West Syndrome: Consensus Statement of the West Delphi Group
Article first published online: 27 OCT 2004
Volume 45, Issue 11, pages 1416–1428, November 2004
How to Cite
Lux, A. L. and Osborne, J. P. (2004), A Proposal for Case Definitions and Outcome Measures in Studies of Infantile Spasms and West Syndrome: Consensus Statement of the West Delphi Group. Epilepsia, 45: 1416–1428. doi: 10.1111/j.0013-9580.2004.02404.x
Andrew L. Lux and John P. Osborne, Bath Unit for Research in Paediatrics, Bath, England, and School for Health, University of Bath, Bath, U.K.; Giuliano Avanzini and Tiziana Granata, Istituto Neurologico “C. Besta,” Milan, Italy; Tallie Z. Baram, University of California at Irvine, Irvine, CA, U.S.A.; Roberto Caraballo, Hospital Nacional de Pediatria “Juan P. Garrahan,” Buenos Aires, Argentina; Kevin Farrell, University of British Columbia, Vancouver, British Columbia, Canada; Ellie Hancock, Great Ormond Street Hospital, London, U.K.; Masatoshi Ito, Shiga Medical Center for Children, Moriyama, Japan; Jun Kohyama, Tokyo Medical and Dental University, Tokyo, Japan; Roshan Koul, Sultan Qaboosh Hospital, Alkhad, Oman; Wei-Ling Lee, National Neuroscience Institute, Singapore; Jianxiang Liao, Shenzhen Children's Hospital, Shenzhen, China; Marissa B. Lukban and Aida R. Mendoza-Salonga, University of the Philippines College of Medicine, Manila, Philippines; Anthony G. Marson, University of Liverpool, Liverpool, U.K.; Brian Neville, Institute of Child Health, London, U.K.; Harry O. Nomayo, Klinikum Weiden Kinderklinik, Weiden, Germany; Hirokazu Oguni. Tokyo Women's Medical University, Tokyo, Japan; Shunsuke Ohtahara, Okayama University Medical School, Okayama, Japan; Hian-Tat Ong, National University Hospital, Singapore; Desiderio Pozo Lauzan, William Soler Children's Hospital, Havana, Cuba; Dietz Rating, University of Heidelberg, Heidelberg, Germany; Bernhard Schmitt, University Children's Hospital, Zurich, Switzerland; Malinee A. Thambyayah, Hospital Selayang, Kuala Lumpur, Malaysia; Federico Vigevano, Ospedale Pediatrico Bambinu Gesu, Rome, Italy; Anannit Visudtibhan, Mahidol University, Bangkok, Thailand; William Whitehouse, University of Nottingham, Nottingham, U.K.; Virginia Wong, The University of Hong Kong, Hong Kong SAR, China; Hideo Yamanouchi Dokkyo University School of Medicine, Tochigi, Japan; Chainllie Young, Washington University School of Medicine, St. Louis, Mo, U.S.A.; and Zhongshou Zhou, China-Japan Friendship Hospital, Beijing, China.
- Issue published online: 27 OCT 2004
- Article first published online: 27 OCT 2004
- Accepted June 20, 2004.
- Infantile spasms;
- West syndrome;
- Clinical trials
Summary: Purpose: To reach a broad consensus on case definitions, outcomes, and outcome measures that will ease future study design and facilitate comparison of data from different studies of infantile spasms and West syndrome.
Methods: Persons who had recently presented or published first-author original research in this field were invited to participate in an e-mail Delphi process and to invite other investigators or clinicians who they thought might participate.
Results: The process consisted of six rounds, anonymous except to the facilitator. In total, responses were received from 46 participants. The final statement was approved by 31 participants from 15 countries. It concluded that the primary clinical outcome, cessation of spasms, should denote absence of witnessed spasms from within 14 days of commencement of treatment, and for ≥28 consecutive days from the last witnessed spasm. Primary electroclinical outcome denotes cessation of spasms with resolution of hypsarrhythmia. West syndrome should be a defined subset of the syndrome of infantile spasms. An infantile spasms single-spasm variant should be recognized. Ways are suggested of handling subtle spasms in the context of clinical studies. It proposes a standard for reporting modifying and atypical features of hypsarrhythmia, a minimal set of baseline characteristics and outcomes that should be reported in trials of infantile spasms, and suggests a standard definition of relapse. Consensus was not reached on a definition of hypsarrhythmia.
Conclusions: We reached a clear consensus on many aspects of study design for the investigation of infantile spasms, although incomplete consensus was found on how to define EEG criteria.
Infantile spasms (West syndrome) is a form of epilepsy that is associated with many underlying conditions and often a poor developmental outcome. Its main clinical features are spasms that usually occur in clusters and onset that almost always occurs during the first 2 years of life. It is characteristically associated with an EEG pattern called hypsarrhythmia, although hypsarrhythmia may not be present throughout the course of the spasms and is not regarded as essential for the diagnosis of infantile spasms. Hypsarrhythmia is an interictal pattern that usually changes during clinical attacks to lower-amplitude slow waves or to a sudden flattening that is known as an electrodecremental period. Various EEG features have traditionally been labeled modified or atypical hypsarrhythmia (1–4).
Several International League Against Epilepsy (ILAE) commissions and workshops have proposed definitions and classifications of infantile spasms, and some elements of these are not mutually consistent. For example, proposals from the ILAE Commissions on Classification and Terminology in 1985 and 1989 suggested that cases of infantile spasms should always have onset of symptoms before age 12 months, whereas a 1991 workshop of the ILAE Commission on Pediatric Epilepsy suggested that infantile spasms “transcends age groups and may occur in infancy or childhood” (5–7). Variation in age at onset as a defining criterion alters epidemiologic characteristics such as median age at onset of illness and affects estimates of age-dependent outcomes. In addition to differences in definitions and criteria proposed by the ILAE, many other variations in definition, classification, and outcome criteria are found in published studies.
An initial proposal for standardized case definitions, outcomes, and outcome measures in infantile spasms was presented at the Symposium on West Syndrome and other Infantile Epileptic Encephalopathies, Tokyo, February 2001 (8). Many participants were enthusiastic about this proposal, although some thought that, in particular, it would be difficult to achieve a consensus on EEG characteristics.
West Delphi followed the format of a Delphi process, a method for eliciting expert consensus developed by the RAND Corporation in the early 1950s (9). A Delphi process essentially involves isolated experts giving their judgments or opinions to a moderator or facilitator, who makes those opinions anonymous and redistributes them to the full group of experts (10). Because the experts are able to relate their opinions to that of the larger group, this facilitates an iterative process by which they can finally reach broad agreement. This method has been used before in the fields of neurology, pediatrics, and psychiatry (11–14). In West Delphi, responses were collated by A.L.L., who acted as facilitator of the process. J.P.O. acted as guarantor of the process but was not aware of the identity of individual respondents and participated as an opinion giver.
Invitations were sent by e-mail to 133 persons who had published articles as primary authors on infantile spasms during the previous 10 years, or who were known to A.L.L. or J.P.O. to have an interest in infantile spasms. Invited persons were asked to forward the e-mail to anyone who they thought would be interested in participating. The first round included 10 questions with multiple-choice format responses (MCQs) that acted to stimulate debate and explored some controversial areas. Those agreeing to participate were asked to formulate further questions as a unique contribution to the study. We made provision for participants to enter after the first round because not all interested persons would necessarily have been contacted in time to respond to Round 1. We suggested that, to be considered members of the West Delphi Group, participants would need to make contributions to other rounds and to agree the final draft of the proposal. We also suggested that minority and dissenting views would be clearly represented in the final draft, and that we would not censor views.
The content and form of subsequent rounds was determined in part by the suggestions of participants. A balance was reached between eliciting quantitative information about the strength of agreements and eliciting qualitative and creative views on how we might approach problems of definition, classification, and measurement. We tried to maintain a focus on aspects of definition and outcome and to develop proposals that would clarify and simplify study design without introducing unnecessary constraints. The first round of West Delphi was sent in January 2001, and the final round was sent in November 2003.
In Round 1, 31 responses were received from participants in 14 countries, and overall, 42 respondents from 16 countries. In addition, three potential participants stated a firm intention to participate but made no substantial contribution to the process. One response was initially on behalf of members of the Königsteiner Kreis, a group of pediatric neurologists that had already started to address the issue of case definitions and outcome measures in infantile spasms. These participants continued to confer and send a single response to subsequent rounds. The 10 initial questions covered the following areas: clinical features, restrictions for age at onset, assessing developmental delay, the usefulness of etiologic subgroups, the likelihood of reaching consensus on a definition of hypsarrhythmia, using aggregated data, defining developmental delay, and the interpretation of adverse effects.
Twenty-seven responses to Round 2 were received. The first section, consisting of seven questions, sought qualitative comments and suggestions about the following: a general definition of infantile spasms, terms for specific clinical and electroclinical scenarios, suggested classifications for etiologic subgroups, necessary and sufficient EEG features of hypsarrhythmia, arguments for and against a 28-day spasm-free period to define the clinical response cessation of spasms (an outcome that had been suggested in Round 1), and a list of appropriate tests to measure developmental outcome. The second section took MCQ format and covered such areas as: clustering of spasms; ictal units; age-related classifications and misclassification of cases; age correction for gestational age at birth; developmental delay at time of diagnosis; use of the terms symptomatic, cryptogenic, and idiopathic; distinguishing spasms from myoclonus; the role of the cortex and subcortex in the etiology of infantile spasms; the potential role of the ILAE-proposed five-axis diagnostic scheme; clinical and electroclinical outcomes; features that constitute relapse; periods of necessary and sufficient therapeutic intervention; and the most appropriate ages for developmental assessment.
In Round 3, participants were invited to respond to 32 statements that represented majority opinions from earlier rounds and to state whether they agreed, were unsure, or would prefer an amendment. They also were asked to state whether they thought the statement should be included in the final West Delphi proposal. The first two statements included a diagram illustrating putative definitional relations between epileptic spasms, infantile spasms, WS, clustered and nonclustered spasms, and hypsarrhythmia. Twenty participants responded. An area identified as important in Round 2 was the status of pyridoxine administration as a diagnostic tool, and discussion of this issue also formed part of Round 3.
Round 4 presented modified statements, requesting comments on their content and suitability for the proposal. Twenty-two responses were received. Five participants were essentially prepared to accept Round 4 statements as the consensus proposal. Round 4 proposed new terms, predisposed and nonpredisposed, in an attempt to clarify inconsistencies in the use of the terms cryptogenic, idiopathic, and symptomatic. Although some participants were enthusiastic about these new terms, they did not meet with universal approval and were dropped. At the suggestion of two participants, an opinion was sought from an external expert on the role that might be played by pyridoxine.
Round 5 consisted of a draft paper for submission as a final proposal from the West Delphi Group. It contained a cover letter outlining areas of incomplete agreement and contention. Before sending this draft, the facilitator (A.L.L.) corresponded independently with several participants to clarify their views as expressed in Round 4. Most participants agreed with the statement as worded at this stage, but 12 participants had substantial comments that were addressed individually. These were reviewed by the factilitator (A.L.L.) and guarantor (J.P.O.) of the study, and comments were sent to the individual participants for clarification or further comment before they were sent to the whole group with a final draft of the consensus document.
Round 6 consisted of a final approval of the draft paper. The detailed responses from each participant remain pseudo-anonymous. That is, they are known to the facilitator but not to any other participant, including J.P.O. The final consensus is printed later. Paragraph numbering is provided so that future investigators may make more specific reference to the proposals.
West Delphi Proposal
1. Introduction to West Delphi. The West Delphi Proposal is a consensus statement that proposes a series of definitions and standards for use in studies of infantile spasms. In particular, it aims to provide clearer definitions of cases and etiologic subgroups, and a standard for outcomes and outcome measures. We hope that this will help those who wish to design future studies of infantile spasms, and that such studies will provide optimal information for those who wish to learn from these studies and to choose the best approach to treating infantile spasms.
2. Infantile spasms. We propose using the term infantilespasms (ISs) to describe an epilepsy syndrome that rarely has onset in children older than 2 years and usually has onset in children younger than 1 year. Its main clinical manifestation is clinical spasms that usually occur in clusters. Many potential etiologies or associated conditions exist. The most characteristic EEG finding is hypsarrhythmia. However, hypsarrhythmia is not found in all cases, nor is it found throughout the clinical course of the condition. Hypsarrhythmia is usually interrupted during a clinical attack of epileptic spasms. The spasms are often associated with developmental arrest or regression.
3. West syndrome. We propose using the term West syndrome (WS) to describe the combination of spasms that occur in clusters and hypsarrhythmia on an EEG. We do not require, as have some previous definitions of WS, that evidence of delayed development occur before the onset of spasms. Some participants wished to classify children with single spasms (that is, without spasms in clusters) and with hypsarrhythmia as having WS, but most preferred to reserve the term West syndrome for cases with clustered spasms.
4. Significance of developmental delay. We see no reason for including developmental delay at the time of onset as a defining feature of WS. Indeed, developmental delay at onset of spasms is likely to be assessed unreliably because spasms are often subtle, and their true time of onset is often unrecognized, and also because development is hard to assess in early infancy. However, we believe that information about unequivocally normal development at the time of onset of spasms is worth recording, because this may be associated with better developmental outcome. Apparently abnormal development may be confounded by the effect of unrecognized spasms at onset or the epileptic encephalopathy.
5. Infantile spasms—single-spasm variant. It is less usual for ISs always to occur singly rather than in clusters, yet we agree that such cases do occur. We propose that these cases are distinguished in studies of ISs by being called infantile spasms single-spasm variant (ISSV). A child with single spasms and hypsarrhythmia should be classified as having ISSV rather than WS. We do not propose a strict definition of clustering, but, as a general guide and until further evidence is available, we suggest that a spasm can be regarded as a single spasm if no other spasms occur for 1 min beforehand and for 1 min afterward. Any definition of clustering should be clearly stated. We wish to emphasize that a child with single spasms and without hypsarrhythmia is rare, and such cases are likely to have a diagnosis other than ISSV. Some participants thought that they would never diagnose ISs in such a case. We would suggest that authors state clearly, in cases with neither clustering of spasms nor hypsarrhythmia, why they are accepting the diagnosis of ISSV.
6. Hypsarrhythmia without infantile spasms. Some children are investigated for symptoms other than clinical spasms and are found to have hypsarrhythmia. If these children, even after video-EEG investigation, are found not to have any evidence of clinical spasms, we propose that they be classified as having hypsarrhythmia without infantile spasms (HWIS). Future studies will inform us whether these children have a high risk of developing ISs at a later date and about their developmental outcome.
7. Relation between definitional terms. We propose using ISs as an inclusive term for all children with clinical spasms who have evidence of EEG abnormalities consistent with the clinical syndrome of ISs—typically, although not necessarily, hypsarrhythmia or modified hypsarrhythmia (see statements 16 and 17)—provided that the EEG findings do not suggest another specific diagnosis. The terms West syndrome and infantile spasms single-spasm variant describe specific subgroups of ISs. Investigators might wish to report studies of children with all forms of ISs and report the proportions in their study with WS and ISSV. Or they might wish to perform a study that included, say, only children with WS and ISSV with hypsarrhythmia. Studies that focus on developmental and EEG outcomes might wish to include children with HWIS (although HWIS is not a subgroup of ISs). Consistent use of these definitional terms and a clear statement of how many children came from each group will assist those who wish to use and interpret study data. We believe that these definitions clearly distinguish clinical subgroups, encourage clarity, and permit flexibility in study design.
8. Our proposed relations between ISs, WS, ISSV and HWIS are illustrated in Figures 1 and 2. These figures also illustrate the relations between these terms and the five-axis system proposed by the ILAE Task Force on Classification and Terminology, 2001 (15).
9. Features of clinical spasms. We propose that the term clinical spasms be used to describe brief and synchronous movements of the head, trunk, and limbs, or sometimes of the head, trunk, or limbs alone. The movements may be flexor or extensor, or a mixture of extensor and flexor movements, and they may be asymmetrical. These movements have a longer duration than the movements of myoclonus, and they are shorter than the movements associated with a tonic seizure. Thus their duration is ∼1 s. The subtlest movement that might be classifiable as a clinical spasm is a head nod. Although it has been suggested that the term describing this seizure semiology should be epileptic spasm (ILAE Glossary of Descriptive Terminology for Ictal Semiology, 2001: paragraph II 18.104.22.168), we propose using the term clinical spasms to describe the ictal phenomenology and reserving the term epileptic spasms to describe the seizure type (16).
10. Subtle spasms. More subtle movements, often described as subtle spasms, may constitute the clinical attack associated with hypsarrhythmia. Such movements include episodes of yawning, gasping, facial grimacing, isolated eye movements, and transient focal motor activity. These patients may be having focal seizures, and the episodes are difficult to count and record. Ideally, a video-EEG would be performed to confirm that clinical spasms are not occurring.
11. Subtle spasms at onset or at follow-up. We propose that children with these subtle ictal events be classified at onset as having hypsarrhythmia without infantile spasms(HWIS) because the events are different from clinical spasms, they are difficult to recognize and count, and they are not reliably measured as a primary clinical outcome. However, we suggest that investigators should, during follow-up of a case of ISs, report subtle movements associated with typical electrodecremental EEG changes or interruption in the interictal background, and that this combination of clinical and neurophysiologic findings would be sufficient to constitute electroclinical nonresponse or relapse.
12. Epileptic spasms. We propose using the term epileptic spasms to describe the epileptic seizure–type of clinical spasms associated with an epileptiform EEG. This corresponds to Axis 2 of the proposed ILAE scheme. This term would include conditions such as the periodic spasms of Gobbi and ISs (17,18). The term epileptic spasms implies that the EEG is consistent with the diagnosis of an epilepsy, but does not by itself imply hypsarrhythmia or any more specific EEG pattern.
13. Relation to proposed ILAE five-axis system. By using these definitions and classification, clinical spasms, epileptic spasms, and ISs correspond with axes 1 (ictal phenomenology), 2 (seizure type), and 3 (syndrome), respectively, of the proposed ILAE five-axis scheme. West syndrome is a syndrome in its own right and a subgroup of ISs, as is ISSV.
14. With respect to Axis 3 (syndrome) of the proposed ILAE diagnostic scheme, we propose that ISs is the syndrome of epileptic spasms with onset generally during the first 2 years of life, in association with high-voltage, multifocal, or diffuse interictal EEG abnormalities (17). We propose that WS be defined by the association of clinical spasms, at least some of which occur in clusters, with an interictal EEG showing hypsarrhythmia. This may be either “classic” hypsarrhythmia, or hypsarrhythmia with modified features. Infantile Spasms single-spasm variant (ISs) is another form of ISs and also is regarded as an epilepsy syndrome.
15. Ictal unit of infantile spasms. Disagreement exists about whether the appropriate ictal unit in studies of ISs should be the spasm or the cluster of spasms. Most West Delphi participants thought that the more appropriate ictal unit was the cluster of spasms. However, we thought that limited value existed in outcomes that rely on measurement of ictal units, such as time to 50% reduction of spasms or the proportion of study participants having a ≥50% reduction in spasms.
16. Hypsarrhythmia. We propose that the term hypsarrhythmia be used to describe an EEG pattern that is characterised by random, high-voltage spikes and slow waves. The most striking features of hypsarrhythmia are high-voltage (generally >200 μV) slow waves with variable amplitude; spikes and waves from many foci, and varying with time; and a lack of synchrony, with a generally “chaotic” appearance. The typical appearance is more likely to be found in earlier stages of ISs and when onset occurs at a younger age. The hypsarrhythmic pattern may disappear during rapid-eye-movement (REM) sleep, but it may be found with greater sensitivity in some other stages of sleep (19).
17. Modified and atypical hypsarrhythmia. With respect to the use of the term modified hypsarrhythmia (1,4) and its synonym (or at least near-synonym) atypical hypsarrhythmia (3), we endorse the recommendation of the 1991 workshop on infantile spasms (7) that study participants with modified hypsarrhythmia not be reported in dichotomized groups but that, where appropriate, a description be given of the individual modifying features. The presence of modified features may depend on the stage of ISs at which the EEG is performed; it may depend on treatment; and as an aggregate variable, it probably has little practical prognostic significance in randomized studies.
18. Forms of EEG recording. EEG findings provide essential information in studies of ISs. We recommend a degree of standardization in the timing of EEG investigations, and that investigators limit potential information bias in studies by making allowance for differences in the sensitivity of different investigations. Increased sensitivity is likely in the case of video-EEGs, with longer periods of recording, and with recordings that include periods of non-REM sleep. Although performing less-sensitive investigations is reasonable, it is prudent to design studies that use more-sensitive investigations before inferring that EEG abnormalities are genuinely not present.
Etiology of infantile spasms
19. Etiologic subgroups of ISs. Several terms have been used to describe etiologic subgroups of ISs, but these terms have not been used consistently between studies. Incomplete agreement remains about the use of the terms symptomatic, idiopathic, and cryptogenic, even though such terms also are used in the classification of other epilepsies. The terms idiopathic and cryptogenic have been used in some studies as synonyms of each other and as antonyms of symptomatic.
20. Idiopathic infantile spasms. The term idiopathic infantile spasms is used to describe cases in which ISs occur without any identifiable underlying cause, other neurologic signs, or symptoms. Such cases may be associated with a family history. Although we agree that cases in this group may have good developmental outcomes, we do not approve making eventual normal development part of its definition, because we are reserving its use as part of an etiologic classification for use at the time of study enrolment.
21. Cryptogenic infantile spasms. The term cryptogenic infantile spasms has been used to describe patients suspected of being symptomatic but for whom an underlying structural or biochemical cause could not be identified. Features suggesting an unidentified underlying cause include preceding developmental delay, or other neurologic features, such as seizures. It has been suggested that the term probably symptomatic is preferred to cryptogenic (15), although West Delphi participants had a range of views on this. Some prefer to retain the term cryptogenic; some prefer probably symptomatic; and some suggest that, if we cannot identify the underlying cause, we ought to use the term nonsymptomatic rather than subjectively to classify cases further.
22. Symptomatic infantile spasms (ISs). The term symptomatic ISs has been used to refer to cases in which the seizures result from an identifiable cause, and often to refer also to cases in which neurologic features or an unequivocal developmental delay precedes the onset of spasms. We suggest reserving the term symptomatic for cases with an identified underlying disorder, and classifying cases with neurologic symptoms, signs, or developmental delay, but no proven cause or etiology, as cryptogenic.
23. Specific etiologies. However constituted, the etiologic subgroups discussed earlier will include children with different underlying causes, which may or may not be identified, and such aggregation into etiologic subgroups is of limited value. However, within any one study, it is possible to obtain precise estimates of effect, and to perform formal statistical tests for heterogeneity of effect, on groups with specific diagnoses only if they are relatively commonly associated with ISs. An example of such commonly associated diagnoses is tuberous sclerosis. Greater homogeneity of study design will help in aggregating data about rare causes of ISs.
24. Timing of etiologic classification. It is often unclear in study reports whether the classification into etiologic subgroups is based on diagnostic information known at the time of treatment decision or randomization, or information available at the time of study completion. If the data are to be used to inform future treatment choice, it is rational for the study to report outcomes according the classification made at the time of treatment decision. This avoids the information bias introduced by reclassification of cases between time of treatment decision and study completion. Initial classification is the natural choice for stratified analyses by intention-to-treat.
25. We ought also to use the more complete and specific diagnostic information that is available at the time of study completion. This informs us about outcomes and treatment choice for children with specific diagnoses, although we should be aware of the bias that might be introduced by incomplete diagnostic information at the time of treatment decision. We propose that, to make this clear, studies reserve the terms cryptogenic, idiopathic, and symptomatic to refer to an etiologic classification made at the time of randomization or treatment choice, and that they report the final diagnostic classification in more-specific diagnostic groups. We suggest a list of specific diagnostic groups in Appendix B. Of course, study-inclusion criteria or stratified randomization might include more-specific diagnostic groups, such as tuberous sclerosis, but the study report should make clear whether diagnostic reclassification of cases has occurred by the time of study completion.
Reporting baseline characteristics in studies of infantile spasms
26. Baseline characteristics in interventional studies. We propose that the following baseline characteristics should be reported for each treatment or intervention group: (a) etiologic group, if used for prerandomization stratification; (b) sex; (c) age at onset (median and range, and proportions in each age category; see statement 27); (d) gestation (proportion born preterm, and median gestation); (e) birthweight (median birthweight and proportion <1,500 g); (f) preceding seizures; (g) concurrent and previous treatment with AEDs or steroids; (h) lead time from apparent onset of spasms to treatment (median and range); (i) presence of hypsarrhythmia; and (j) unequivocally normal development at apparent onset of spasms. Attention should be drawn to large differences that might bias the results, even where treatments have been allocated randomly and these differences must have occurred by chance.
27. Classification by age at onset of spasms. ISs typically have an onset between ages 3 and 12 months. Younger and older children are more likely to have other conditions, and a greater risk of misclassification exists. Age at onset of spasms also may influence outcomes. We propose that studies of ISs report outcomes for children in age at onset groups: (a) younger than 3 months (corrected for preterm delivery); (b) 3 months (corrected age) and up to 12 months; (c) 12 months and older. We propose that these be referred to by the terms: early onset, classic onset, and late onset, respectively. Onset would rarely occur after age 2 years, but it is biologically implausible to set an absolute upper age limit.
Duration of treatment
28. Duration of treatment in interventional studies. We propose that treatments for ISs should be considered to have had a sufficient therapeutic trial after 14 days. In general, it is necessary to give a treatment for 14 days before deciding that it has no treatment effect, but discontinuation should be considered sooner if an adverse effect of that treatment appears. We suggest that any study that uses a different period of necessary and sufficient therapeutic trial should state a rationale or empirical evidence for that decision. If adjunct treatments are used, the study protocol should state clearly how these might affect outcome.
29. Primary clinical outcome. We propose that the primary clinical outcome of studies of ISs is cessation of spasms, and we suggest a standard definition for this outcome. Without qualification, the term cessation of spasms should denote that no clinical spasms have been witnessed from a time commencing within 14 days of treatment, and for a period of ≥28 consecutive days from the time of the last witnessed spasm. Studies of ISs should report the number and proportion of children in each treatment group who meet these criteria.
30. Timing of EEG investigations. Although we do not have evidence for the clinical importance of cessation of spasms and resolution of abnormal EEG findings, in terms of prognosis for development and function, we propose that it is essential to report the EEG status of children at baseline and at follow-up. We suggest that an EEG be performed at least:
- • before treatment is allocated;
- • between study days 14 and 21 (that is, within 7 days of the end of the period of sufficient therapeutic trial); and
- • between study days 42 and 49 (that is, after the period during which 28 days of freedom from spasms should have occurred).
31. EEG classification.Resolution of hypsarrhythmia is a more appropriate EEG outcome than normalization of EEG, because with many underlying etiologies, we cannot expect the background EEG to normalize even if epileptiform changes resolve. Progression to other seizure types also may occur without any evidence of ISs or hypsarrhythmia.
32. If less-sensitive investigations do not show hypsarrhythmia, we recommend that a sleep EEG be performed to provide stronger evidence that the hypsarrhythmia has resolved. Ideally, a video-EEG recording would include sleep. No evidence suggests recommending a specific duration for video-EEG recording, but a 24-h period is typical, and some participants thought 6 h would be as sensitive as 24 h. Although it might be inferred that the EEG remains abnormal in cases in which continuing spasms are found, it also is useful to have information from EEGs in children with continuing spasms. We regard this information as essential because it will help to inform us that the seizure type has not changed, about the prognostic importance of EEG findings, and also about treatment effects on the EEG.
33. Primary electroclinical response. Some West Delphi participants are uneasy with the idea that children with no witnessed spasms but continuing hypsarrhythmia should be classified as primary responders. They suggest that a primary responder should have both cessation of spasms and resolution of hypsarrhythmia. We propose the term primary electroclinical outcome to describe this combination of outcomes, and we propose that it be regarded as an essential outcome (20).
34. Studies that report both primary clinical outcome and primary electroclinical outcome provide important information about how often the clinical response is associated with resolution of hypsarrhythmia. Such studies would permit a degree of extrapolation of data from studies performed in places where access to neurophysiologic equipment is more restricted and where the primary electroclinical outcome cannot be reliably determined.
35. Deaths during study period. Studies should report all deaths occurring during the study period, even when they are not regarded as attributable to treatment. Study reports should state which adverse events were investigated, how they were sought, and how they were classified.
Outcome measures and analysis of data
36. We propose that at least the following outcomes and measures, stratified by intervention group, be reported in studies of ISs:
- • primary clinical response
- • primary electroclinical response
- • relapse-free primary responses (clinical and electroclinical, numbers and proportions)
- • continuing subtle spasms where clinical spasms have ceased
- • distribution of time to relapse (from time of randomization, perhaps best represented graphically as a variable proportion with relapse-free remission over time)
- • development at age 2 years (medians and ranges for tests chosen)
- • deaths and other serious adverse events (numbers and descriptions of events)
- • presence of and progression to other seizure types (numbers, proportions, and seizure types)
- • nonserious adverse events (numbers, proportions, and description of events)
37. Analyzing different etiologic subgroups. We suggest that most outcomes are influenced more strongly by underlying disorders than by treatment. We recommend that investigators exercise caution in performing multiple tests on such outcomes and attributing significant differences to treatments because, in addition to there being biases introduced by multiple testing, the conclusions may lack biologic plausibility. However, one aim of treatment is to improve development and function by limiting any damage that might be caused by repeated spasms.
38. Analyzing time to cessation of spasms. Time to cessation of spasms can be represented by actuarial curves and is a clinically useful outcome that might inform future study designs. We suggest the convention of reporting time to 28 days of spasm freedom for the primary clinical response. This is a useful and intuitive clinical measure that can be examined by using the statistical techniques of survival analysis. Measures of time to cessation of spasms are more reliable and informative than measures of time to 50% reduction in ictal units. The primary electroclinical response does not lend itself so readily to such analyses because EEGs are performed in a cross-sectional fashion, whereas clinical observation is a more continuous measure.
39. Relapse. The following should be considered to constitute a clinical relapse of ISs at any stage after a primary clinical response has been obtained: (a) any episode of spasms occurring in clusters; (b) two or more episodes of spasms that occur singly but not in clusters; and (c) subtle spasms (if accompanied by an EEG showing appropriate changes). A single witnessed spasm would not be reliable and should not be classified as a relapse of clinical spasms, but investigators might wish to report their occurrence.
Role of pyridoxine (vitamin B6)
40. Diagnostic role of pyridoxine. Pyridoxine-dependent seizures are rare (21). They even more rarely have the clinical manifestation of infantile spasms, although this is thought to be most likely when other seizure types have occurred before the onset of spasms. Giving intravenous pyridoxine is associated with a risk of apnea. It requires close monitoring and is not known to be associated with rapid resolution of hypsarrhythmia. Oral pyridoxine seems to be associated with a median time to response of several days. Thus incorporating a diagnostic and therapeutic test of pyridoxine into study designs is challenging.
41. Treatment role of pyridoxine. In addition to use in diagnosing potential pyridoxine-dependent seizures, pyridoxine has been used more broadly as a treatment or adjunct treatment for ISs. The Cochrane Review of treatment of ISs did not identify any evidence of treatment effect in trials meeting its inclusion criteria (22,23). If pyridoxine is used, the report should clearly state its timing, dose, duration, and whether it is being used to exclude a diagnosis of pyridoxine-dependent seizures or as a treatment intervention for ISs in its own right.
Areas of incomplete consensus
42. Areas of incomplete consensus remain. In the future, we might wish to develop a more detailed structure for reporting adverse events and recommendations on which tools to use to assess developmental outcomes and to classify disability. Finally, we found it difficult to provide a detailed description of the limits of definition of hypsarrhythmia: those designing studies must consider the advantages of using several EEG assessors, and how to assess intrarater and interrater agreement in classification of EEGs.
The rationale behind West Delphi was to propose standard case definitions, outcomes, and outcome measures that will ease future study design; that facilitate comparison of data from different studies; and that encourage studies to answer questions that are important to clinicians and families seeking information about treatment choices. These aims are essential stages on the road to determining the best treatments for children with ISs. The process dissected many of the issues surrounding case definitions and outcomes. Although substantial areas of consensus exist, several areas require further discussion. We wish to emphasize that the primary aim of the consensus is to aid the design of studies of ISs.
Case definitions for studies of infantile spasms
We aimed to propose definitions that are congruent with previous usage and recent ILAE proposals. For example, Roger and Dulac (17) used the terms infantile spasms and West syndrome to refer to syndrome diagnoses, and they suggested the term epileptic spasms to refer to a seizure type. The West Delphi consensus evolved to have strong congruence with these definitions.
However, we were prepared to make clear distinctions and depart from previous practice where we thought that this would be valid and useful. For example, we found it valuable to use the term epileptic spasms to describe the seizure type, rather than as a phenomenologic description, as proposed in the ILAE Glossary of Descriptive Terminology, 2001 (16). Where the ILAE Proposed Diagnostic Scheme for People with Epileptic Seizures and with Epilepsy suggests the term spasms to describe the seizure type, we propose making a distinction between clinical spasms, which describe seizure phenomenology, and epileptic spasms, the seizure type. We think that these terms are more appropriate and that this use of the modifiers clinical and epileptic has strong intuitive appeal.
We generally found the proposed five-axis classification of epilepsy, the overall diagnostic scheme of which was proposed by the ILAE General Assembly in May 2001, to be potentially useful, and we described where our classification is congruent with that classification. We wait to see if the details of those proposals will find broad approval, and we suggest the changes we outline as modifications to the scheme (15).
It seems that nobody would dispute the status of West syndrome as an epilepsy syndrome, but initial disagreement was found about the status of developmental delay as a defining feature of WS. The consensus was that it is often difficult to determine when spasms started and whether development was normal at that stage. Because these are unreliably determined, and misclassification is likely, we thought that developmental delay at onset of spasms ought not to be a defining feature of WS. It was suggested that WS is defined as a new triad: spasms, clusters and hypsarrhythmia. However, a consensus formed that, even though it might make a useful but informal aide mémoire, this triadic definition was too contrived to be adopted formally.
We debated the status of infantile spasms as a syndrome, and also the status of infantile spasms single-spasm variant with hypsarrhythmia. An argument was reached for classifying these as seizure types, corresponding to Axis 2 of the proposed ILAE diagnostic scheme. However, we did not find this argument compelling and thought that it is more appropriate to classify ISs and ISSV with hypsarrhythmia as epilepsy syndromes. The features that define WS as a syndrome subgroup of ISs are clustering of spasms and hypsarrhythmia. Hypsarrhythmia remains rather loosely defined and has traditionally been prone to reclassification when modifying features exist. Even the status of a criterion to define nonclustering of spasms requires empirical clarification, because at least one contributor thought that spasms separated by longer than 1 min might genuinely be part of a single cluster.
As was predicted by several respondents to the paper presented at the International Symposium on West Syndrome and Other Infantile Epileptic Encephalopathies, Tokyo, February 2001, the task of defining necessary and sufficient features of hypsarrhythmia proved very challenging. We eventually agreed on a description of hypsarrhythmia rather than a strict definition of limits, and evidently an element of Gestalt remains in recognizing this EEG pattern. We are not aware of studies of intrarater and interrater agreement with classification of EEGs as hypsarrhythmic. Such studies would inform future study design and, we hope, would provide reassurance that misclassification of EEG patterns is not a substantial problem. Meanwhile, those designing studies might consider reducing bias and misclassification by, for example, using EEG assessment by two independent, treatment-blinded investigators.
We have followed the proposal made by the 1991 ILAE Workshop in recommending that modifying features be listed separately rather than regarded in the aggregate as forming a distinct group of EEG types (7). Individual modifying features are listed in Appendix A. Modified EEGs may bear more or less resemblance to “typical” or “classic” hypsarrhythmia, and we think that it is irrational to dichotomize cases on the basis of presence of any one or more modifying features. EEG modification may be related to a missed temporal window of opportunity for catching a “typical” recording, because increased synchrony and organization is related to brain development or sleep state, or it may relate to underlying causes or triggers: for example, increased focality where the trigger is a cortical malformation or tuber.
Although some participants expressed reservations, it was agreed that cases of ISs might occur without hypsarrhythmia. Some studies have reported hypsarrhythmia in fewer than half of cases classified as having ISs at the time of randomized treatment, (24), and it is well recognized that hypsarrhythmia is not found throughout the course of ISs.
With respect to ISSV without hypsarrhythmia, we thought that we could not exclude these solely by definition if we are prepared to admit cases with single spasms with hypsarrhythmia and cases with clustered spasms without hypsarrhythmia. However, we would counsel caution and clarity when including such cases in studies of ISs, and some contributors were certain that they would never diagnose ISs in such cases. Spasms that are never observed to cluster and that are not associated with hypsarrhythmia are unlikely to be ISs, and if such cases are included in studies, we would expect the investigators to state clearly their reasons for regarding these cases as valid.
Some contributors thought that unequivocally normal development at the time of onset of spasms is a good prognostic feature. They suggested that this should be recorded and used for prognostic factor analysis in studies of ISs. It also was suggested that this reframing from apparently delayed development to unequivocally normal development is useful when it comes to counseling families about prognosis.
It was evident from the process of eliciting a consensus that clinicians and investigators have used the terms infantile spasms and West syndrome variably, and that great variation has been noted in the interpretation of the terms cryptogenic, idiopathic, symptomatic, and nonsymptomatic. Previous ILAE commissions have used the terms idiopathic and cryptogenic as antonyms of symptomatic (5,6). Participants debated these terms a great deal. The eventual consensus was that information about prognosis and treatment response related to specific underlying diagnoses is probably more useful than classification into etiologic subgroups that are themselves heterogeneous.
We also discussed the issue of potential information bias related to reclassification of cases between enrolment in a study and the time of study completion. We believe that clearly distinguishing between an etiologic classification made at the time of study enrolment and a diagnostic classification made at the time of study completion will allow those using information from studies to be clearer about what type of diagnostic information tends to be available at the time of treatment decision.
Some participants wished to emphasize the limited value of aggregating cases into etiologic subgroups, preferring to look instead at outcomes in groups of children with the same specific underlying disorder. For example, tuberous sclerosis seems to have a better response to treatment with vigabatrin do than other cases of symptomatic infantile spasms.
One controversial area was that of subtle seizures associated with hypsarrhythmia but without evidence of spasms. Whereas eye or subtle distal limb movements may be regarded as evidence of continual seizure activity, they do not provide a reliable outcome for the purposes of clinical studies because they are likely to be missed without careful observation. In the absence of hypsarrhythmia, these subtle movements might be classified as focal seizures. However, it is important to recognize cases that have interictal hypsarrhythmia and typical ictal EEG changes without clinical spasms. Although we suggest classifying these as HWIS, little doubt exists that some of them still effectively have the electroclinical syndrome of ISs without the typical clinical manifestation. Again, we think that having a specific definition and term to describe such cases will allow their appropriate inclusion in studies without confusing the reader. However, we did think that such subtle movements were sufficient to define electroclinical nonresponse—they would presumably be associated with an EEG correlate; otherwise no evidence would support the idea that such subtle movements were epileptic—or electroclinical relapse in cases in which spasms had ceased.
Outcome measures for studies of infantile spasms
The latter part of the consensus statement (sections 26 onward) addresses the issue of standardized outcomes and outcome measures for studies of ISs. Most published studies have provided incomplete information about even essential outcomes, such as how they defined clinical or electroclinical response. In the early rounds of West Delphi, it became apparent that most clinicians would not regard cases relapsing within a few weeks as primary clinical responders. However, they agreed that spasms recurring after 28 days of absence are described reasonably as relapse rather than as failed primary response.
One question that emerged was that of defining periods that constitute necessary and sufficient durations of treatment. The consensus was that 14 days was a suitable duration for both—at least, with current treatments. In other words, we agreed that a treatment has had sufficient opportunity to prove its effect by 14 days, and that it is necessary to give a treatment for this duration before concluding that it has failed to show any treatment effect.
These limits are arbitrary to some degree, and good reasons might be found for a new study to use a different duration. However, we think it is useful to suggest a standard, and we would invite investigators to explain their reasons for choosing a different period. A danger exists that an extra degree of subjectivity enters study design and makes aggregation and use of information more difficult if, for example, investigators hoped to favor a drug that is reputed to act more quickly than another. We believe that such questions would be better investigated by other means than altering the primary outcome, such as modeling time to cessation of spasms. This would preserve a degree of study standardization that would facilitate comparison and aggregation of data from studies. In the longer term, such an approach would provide better information for clinicians and families.
We included three categories of age at onset (or more correctly apparent onset, because timing can be unreliable) not to define further subgroups but rather to facilitate its use as a prognostic factor.
With respect to secondary outcomes, strong support was found for other recommendations, and in particular, for developmental assessment at age 5 years rather than the more conservative 2 years stated in the consensus. We have limited the statement to what we consider to be a minimal standard of information.
A good deal of negotiation occurred about the most appropriate primary outcome. Most participants were keen on emphasizing the clinical response, but others were sceptical at the idea of suggesting that a response had occurred when the EEG still showed hypsarrhythmia. The final consensus is that two proper primary outcomes are found: the primary clinical outcome and the primary electroclinical outcome, as defined earlier. We think that the approach of reporting both outcomes has many advantages.
Although higher estimates have been made, worldwide probably ∼20,000 to 50,000 cases of infantile spasms are seen each year, and very few of these are enrolled in formal studies (25). Reliably and convincingly to evaluate treatments and to investigate potential variation in responses in different diagnostic groups, we should consider more collaborative and international studies and aggregate data from disparate studies. We hope that West Delphi is a useful contribution to this process.
The clinicians and investigators who contributed to West Delphi have an interest in ISs. The primary interest of many is that of making treatment decisions and using reliable and readily interpretable published information to make those decisions. Many of the contributors also have an interest in designing and contributing to studies of ISs. The process involved novel suggestions and lively debate.
A strict interpretation of the guidelines for a Delphi process would not have permitted conferring by West Delphi contributors who were also members of the Königsteiner Kreis. The Königsteiner Kreis is a group of German and Swiss doctors with an interest in pediatric epilepsy who deferred completion of an ongoing consensus statement while two of its members contributed to West Delphi. One of those members sought the opinions of other members of that group, but the response to West Delphi was counted as a single response to prevent any bias being introduced from coordinated responses.
These proposals are not intended to be immutable guidelines for study design in ISs, but rather to be an aid to encourage the collection of clear, consistent, and reliable information. We would be pleased if they stimulate research that later refines some or all of these proposals. Some of the proposals rest on arbitrary but agreed limits, such as the number of days of spasm freedom that is generally regarded as constituting a primary clinical response. Future studies using these proposed case definitions and outcomes are likely to have wide acceptability and applicability. Inevitably, some participants have accepted, in the cause of consensus, definitions and ideas that would not be their personal first choices.
Studies cannot answer all questions put by all people, and we suggest that these proposals are better regarded as a minimal standard than as sufficient for any study. We hope that constructive criticism of these proposals will lead to refinement and improved consensus, and that it will, in the meantime, provide a framework for the design of informative and valuable studies of ISs.
Acknowledgment: We thank the following: Professor Yukio Fukuyama (Tokyo) for his warm reception of the idea behind West Delphi and the invitation to Professor Osborne to speak at the 2001 Tokyo Symposium on Infantile Spasms and West Syndrome. We thank the members of the Königsteiner Kreis [B. Schmitt (Zürich), U. Brandl (Jena), R. Korinthenberg (Freiburg), I. Krägeloh-Mann (Tübingen), G. Kurlemann (Münster), D. Rating (Heidelberg), M. Wolff (Tübingen)], who had already prepared a consensus paper in German on a similar theme. Two of its members made substantial contributions to West Delphi and drew on ideas that had been developed in the Königsteiner Kreis consensus. We thank Dr. Peter Baxter (Sheffield) for information about the use of pyridoxine; Heather Hill (Bath) for able assistance with secretarial tasks; Cow & Gate, Ltd., Bishopsgate Financial Management, and the Bath Unit for Research in Paediatrics (BURP) for assistance with funding.
Appendix A: List of EEG features that are regarded as modifying features of hypsarrhythmia
These features are traditionally regarded as defining modified hypsarrhythmia, but we recommend listing individual features. One also should note that many of these features are matters of degree rather than absolute differences.
Asymmetry Consistent focal discharge Episodes of voltage attenuation (local, regional, or general) Excessive rapidity Excessive slowing Fragmentation Increased interhemispheric synchronization Increased periodicity Predominant high-voltage, bilaterally asynchronous slow activity
Appendix B: A suggested list of diagnostic categories
This classification uses the 1996 pediatric adaptation of ICD-10.
This grouping specifically excludes transient and specific conditions of the newborn that are classified under perinatal.
|Enzyme deficiencies, including biotinidase, Krabbe, Leigh's, pyridoxine, adrenoleucodystrophy, etc.|
This grouping also excludes conditions mentioned elsewhere, including perinatal.
|Cerebral palsy (unknown cause ONLY)||G80–83|
|Other (ONLY if unable to classify elsewhere; e.g., leucomalacia not due to perinatal cause).|
This grouping also excludes conditions mentioned elsewhere, including perinatal.
|Cerebral infarct or stroke||I63, 64|
|Maternal factors (e.g., drug abuse)||P04|
|Birth trauma, including hypoxic–ischemic encephalopathy (HIE) due to trauma||P10–11|
|Intrauterine asphyxia (HIE)||P20–21|
|Infections: meningitis, cytomegalovirus (CMV), toxoplasmosis, herpes||P35–36|
|Intracranial nontraumatic hemorrhage|
|Transient endocrine or metabolic diseases of the newborn (e.g., hypoglycemia)||P70–74|
|Other: Periventricular hemorrhage/leucomalacia from preterm injury||P91|
|Hypoxic–ischemic injury of uncertain cause|
|Malformations||Agenesis of the corpus callosum||Q 04|
|(if not chromosomal)||Agyria/polygyria|
|Other malformations||Incontinentia pigmenti||Q 82|
|(specific diseases)||Neurofibromatosis||Q 85|
|Tuberous sclerosis||Q 85|
|Hypomelanosis of Ito||Q 92|
|Chromosomal||Down syndrome||Q 90|
|Other: please specify|
Signs and Symptoms not otherwise specified
This section would include information gained from imaging that is not in itself diagnostic, for example:
Cortical atrophy: diffuse Cortical atrophy: focal Hemimegalencephaly
Intracranial injury, for example, due to nonaccidental injury or trauma.
- 1Infantile spasms. In: Atlas of electroencephalography: epilepsy. Cambridge , MA : Addison-Wesley, 1952: 24–30, 52., .
- 2Ictal aspects with emphasis on unusual variants. In: DulacO, ChuganiHT, Dalla BernardinaB, eds. Infantile spasms and West syndrome. London : Saunders, 1994: 36–51..
- 9Experts in uncertainty: opinion and subjective probability in science. New York : Oxford University Press, 1991..
- 10Eliciting and analyzing expert judgment. Philadelphia : Society for Industrial and Applied Mathematics, 2001., .
- 17West syndrome: history and nosology. In: DulacO, ChuganiHT, Dalla BernardinaB, eds. Infantile spasms and West Syndrome. London : WB Saunders, 1994: 6–11., .
- 19variations and pitfalls. In: DulacO, ChuganiHT, Dalla BernardinaB, eds. Infantile spasms and West Syndrome. London : WB Saunders, 1994: 63–81., . :
- 21Pyridoxine dependent and pyridoxine responsive seizures. In: BaxterP, ed. Vitamin responsive conditions in paediatric neurology. London : Mac Keith Press, 2001: 109–65..
- 23Treatment of infantile spasms (Cochrane Review): The Cochrane Library. Oxford : Update Software, 2002., , .